Machines, for example, marine vessels, earth-moving machines, construction equipment, etc. include numerous components that may wear and/or fail over time. Repair or replacement of worn out or failed machine components may require removing the machine from service, which may decrease utilization of the machine. The time required to procure a replacement part, or unavailability of repair personnel or facilities may further increase the down time associated with performing maintenance on the machine. Therefore, it is desirable to monitor the wear rate of machine components and/or to predict when the machine may require maintenance including repair or replacement of worn out components. Such monitoring and prediction capabilities may also allow procurement of replacement parts in advance so that the parts are available when the maintenance is scheduled on the machine. In addition, such monitoring and prediction capabilities may make it possible to schedule maintenance activities during a period when the machine is not expected to be in use, thereby maximizing utilization of the machine.
One approach to monitoring the health of machine components relies on analyzing data collected from various sensors associated with the machine. However, collecting sensor data continuously can be expensive because it may require dedicated systems for collection and storage of the sensor data. On the other hand, periodic data collection may mask trends that may help determine the onset or severity of wear of a machine component. Collecting sensor data periodically may fail to capture sensor data during unexpected events, for example, during an unscheduled shutdown of the machine. Thus, there is a need for a diagnostic model that allows collection of data from the right sensors at the appropriate time.
The model generation and monitoring system of the present disclosure solves one or more of the problems set forth above and/or other problems in the art.